Not Just Men: Fingerprints Suggest Women and Children Helped Build China’s Terracotta Army

There are many archaeological wonders across the world. One of the most unique, in my opinion anyway, has to be the Terracotta Army.

Made up of over 8,000 life-sized figures, mainly soldiers but also acrobats, officials, and musicians, the court of the first emperor has stood watch over his tomb for centuries.

Since their discovery, scholars have been debating these statues: were they made all at once, or was each piece finished by different people and workshops and assembled piece by piece? Were they made by one workshop or multiple? And of course, who made them?

It had long been assumed that the Terracotta Army had been crafted by adult males. But without any records detailing their creation, there is no way to know for sure, or is there?

Terracotta, you see, is a type of clay, so to make each figure, someone had to mould and sculpt it, leaving behind fingerprints. Fingerprints can tell you about someone’s age and sex. Older fingerprint methods, however, have flaws: they treat ridge width and ridge density as two independent variables, but in reality they’re two sides of the same coin. Ridge density is essentially the inverse of ridge width, so using both counts the same evidence twice and makes the results seem more certain than they are.

A new study tackles this problem with a statistical approach: Bayesian mixture models based on ridge width. What they found? It seems the people who made the Terracotta Army were just as varied as the statues themselves.

The Immortal Emperor and his Army

The terracotta army had guarded the tomb of the first emperor for centuries until, in 1974, farmers digging a well in China’s Shaanxi province came upon pieces of a clay figure. These fragments were the first of many clues that eventually led to the discovery of massive chambers filled with life-sized warriors and horses guarding the tomb of Qin Shi Huang. Qin Shi Huang was also called China’s first emperor. He became king of the Qin state at the ripe old age of 13. Then declared himself emperor in 221 BCE.

Over the course of his lifetime, he conquered rival states and unified China.

He is also notorious for his obsession with immortality. As you may have guessed, China is not currently ruled by a centuries-old emperor named Qin Shi Huang, so he failed. Qin Shi Huang died in 210 BCE.

However, about a century later, a scholar wrote that he filled his tomb with rivers of mercury. Today, measurements near his tomb have confirmed high levels of mercury, though no one has dared to open it for various reasons. And my own guess: if mercury levels really are that high, I have a feeling opening a tomb filled with it might be a bit of a health hazard.

For decades, researchers have tried figuring out how the Terracotta Army was made. Inscriptions show that there were two kinds of makers involved: royal potters who marked figures with the character Gong (宫), and local potters.

Some scholars believe separate workshops mass-produced body parts (arms, legs, heads, etc.) and then assembled them piece by piece. Others suppose each workshop made completed statues and transported them to the tomb. Exactly which is true is uncertain, though chemical tests on statues marked Gong (宫) and those marked Xianyang (咸阳) suggest at least two different workshops may have been making the statues.

But beyond that, little is known about the makers. It long was believed that only adult men created the army, although Qin accounts also mention women, young boys, and young girls working in related sectors such as lacquerware.

Clues Imprinted in the Clay

Prior work tried to decipher the maker's sex and age using older methods that look at the width and density of ridges. Additionally, the method cannot distinguish between an adult woman and a teenage boy, because fingerprints shrink when clay is fired, and because populations differ.

That’s where the new study comes in. They decided to try a different approach: a Bayesian mixture model. The researchers started on a small subset of statues, around 227 figures located in Unit T23 of Pit 1. During restoration, the team photographed the prints, choosing only those that were clear, unsmudged, without overlapping prints, and with at least six visible ridges in a row.

That gave them around 91 prints from 36 figures. Then they checked whether the prints might have come from the same person, which would skew the results, and removed those they thought were ‘duplicates’. That left them with 79 prints.

Earlier methods had tried to determine each individual fingerprint’s age and sex using formulas and cutoffs. Basically, something like ridges narrower than X mm belong to a child, and ridges wider than Y mm belong to an adult man. Think of it like this: I could try to classify a bunch of people as male and female based only on their height. A person 4 feet or shorter is a child, and a person 6 feet or taller is male. But what about in between? A tall child can be 5 feet tall, a tall woman can reach over 6 feet, and a short man can be five feet. So those cutoffs can be misleading.

That same problem occurs in fingerprints, with teenage boys and adult women often having overlapping fingerprint metrics.

Thus, the authors state that figuring out the exact sex and age of any given fingerprint is impossible. But figuring out the overall proportion of males, females and children may be more reliable.

Okay, I can barely wrap my head around stats, but I will try my best here. Basically, instead of labelling each fingerprint as definitely male, female, adult or child, the researchers tried to estimate the overall mix of people behind the whole set. First they checked the variability between fingerprints. The spread was wide enough to suggest a mix of males, females and children had made them, so they used a model that allowed for estimates of all ages and both sexes.

Next, they adjusted the ancient measurements to be comparable to modern ones. Why? Remember, fingerprints shrink on clay, and measurements made on modern populations are not shrunk. People in different populations and time periods can also have slightly different finger sizes. Then they ran the Bayesian model. This works by combining existing knowledge (data on modern fingerprints, ages and sex) with new evidence (in this case, the ancient terracotta prints) and then providing probabilities rather than definitive verdicts.

That test is run again and again tens of thousands of times. Each time, the model guesses an age and sex for every print and checks how well those guesses match the real measurements, until it settles on a range of likely mixes of men, women and children that best explains the whole set.

Who Made the Terracotta Army

Okay, we can't say who made the WHOLE Terracotta Army, but we can get a pretty good idea of the proportion of people who made the figures in that section of pit 1.

First, the authors suggest that the spread of prints across the heads, arms, armour, and other parts of the figures supports the idea that the soldiers were made in separate pieces and assembled later. They also suggest that prints on the heads and arms may have been left by workers gripping these parts while joining and smoothing them together.

Second, the prints were likely made by around 56% males and 44% females. This means the army was not made only of men. But what about children?

For this, the researchers also turned to Qin legal documents to see what actually constituted a child. According to the Shuihudi Bamboo Slips (睡虎地秦简), the Rites of the Zhou (周礼) and the Hanshu (汉书). People in the Qin and Han dynasties below the age of 15 were not legally of age. Based on this, around 45% of prints came from adult men, 33% from adult women and 22% from children.

Interestingly, workers seem to have been largely young adults, possibly aged around 20 for men and 18 for women.

But the study also stresses that the model only gives them probabilities of prints. That means 44% female fingerprints does not mean 44% of the workers were female. Additionally, skill can skew who leaves prints; an experienced craftsman leaves fewer prints than an inexperienced apprentice or child. Second, the model assumes the prints of the ancient Chinese workers will match the modern Czech fingerprints on which it is built, a sample with more women, while reference data for East Asian populations is lacking.

Despite these limits, the authors argue that two findings hold up. The fingerprints show a wide variety of ridge widths, too much variety to come from a single age or sex group, and the model consistently indicated that at least some adult women and children were part of the workforce.

But who were these people and why was the workforce so mixed? The researchers point to ancient texts showing that workforces often constituted free peasants, craftsmen, convicts, debtors, and slaves. Women also often fell into these categories and are described working in projects like earth moving. The emperor’s official ideology assigned women clear duties that may have made their involvement in state projects acceptable.

As for children. According to Qin law, it seems child labour was just part of an organised system. According to Shuihudi Bamboo Slips (睡虎地秦简), five young child slaves are equivalent to one craftsman, and Yuelu Academy Qin slips (岳麓书院秦简) describe training periods for apprentices, suggesting state workshops had formal apprenticeships that may account for some of the children working at the Terracotta Army.

The study doesn't give a reason for the male-only assumption; it simply says that is what is usually thought in Qin studies. That isn’t exactly a very scientific approach; it is a very human one, though. If I say nurse, what do you picture? Probably a female. If I say medieval squire, you likely think of a young boy.

Another guess is the stamps identifying the workshops, which in turn would help identify the lead foreman responsible for any given statue. Together they create an image of an all-male workforce.

The study provides the first biological evidence that women and children helped build the Terracotta Army, and that this kind of fingerprint analysis could also reveal who did the work behind other ancient crafts. When so much of history is unknown and built on assumptions, findings like these can illuminate our past and help us better understand the people responsible for creating it.

The Terracotta Army was massive, so I am not surprised that women and children helped, but my lack of surprise is based on the assumption that it had to be so. The fingerprints provide the evidence.

Let me know what you think, and if you'd like to support this blog and the person writing it, you can chip in once, monthly, or yearly. Only if you'd like to.

 References

Li, X., Shen, J., Lan, D., Wang, J., Yu, Y., Cai, J., ... & Zhou, P. (2026). Who produced Emperor Qin Shihuang’s Terracotta Army, China? A Bayesian mixture model approach to fingerprint age and sex determination. Journal of Cultural Heritage, 82, 274-281. https://doi.org/10.1016/j.culher.2026.09.014

Lubow, A., & Anderson, S. (2024, April 19). What you need to know about China’s terra-cotta warriors and the first Qin Emperor. Smithsonian Magazine; Smithsonian Magazine. https://www.smithsonianmag.com/history/what-you-need-know-about-chinas-terra-cotta-warriors-first-qin-emperor-30942673/